Micron’s HBM3E and HBM4 production capacity sold out

Recently, statements from Micron’s CFO regarding HBM capacity and roadmap have garnered significant industry attention.
He revealed that the company’s HBM3E and HBM4 capacity for next year is already fully booked, while confirming that HBM4 will begin shipping in the second quarter of next year.

This situation is not an isolated case; SK Hynix and Samsung’s HBM capacity for 2026 has likewise been largely secured by major customers in advance, reflecting a severe supply-demand imbalance in the global HBM market.
Driven by the generative AI wave, HBM has evolved from a key memory component into a strategic resource influencing the deployment of AI infrastructure.

The root cause of this “capacity rush” lies in HBM’s irreplaceable role as high-performance memory for AI GPUs.
Extremely high technical barriers and complex stacking processes constrain rapid capacity expansion, while market demand is growing exponentially.